Pre-classical, adhoc-but-easy method of
converting an analog prototype filter to a digital IIR
filter. Does not preserve any optimality.
Impulse invariance means that digital filter impulse
response exactly equals samples of the analog prototype impulse
response:
∀n:hn=
h
a
nT
n
h
n
h
a
n
T
How is this done?
The impulse response of a causal, stable analog filter
is simply a sum of decaying exponentials:
H
a
s=
b
0
+
b
1
s+
b
2
s2+...+
b
p
sp1+
a
1
s+
a
2
s2+...+
a
p
sp=
A
1
s-
s
1
+
A
2
s-
s
2
+...+
A
p
s-
s
p
H
a
s
b
0
b
1
s
b
2
s
2
...
b
p
s
p
1
a
1
s
a
2
s
2
...
a
p
s
p
A
1
s
s
1
A
2
s
s
2
...
A
p
s
s
p
which implies
h
a
t=
A
1
ⅇ
s
1
t+
A
2
ⅇ
s
2
t+...+
A
p
ⅇ
s
p
tut
h
a
t
A
1
s
1
t
A
2
s
2
t
...
A
p
s
p
t
u
t
For impulse invariance, we desire
hn=
h
a
nT=
A
1
ⅇ
s
1
nT+
A
2
ⅇ
s
2
nT+...+
A
p
ⅇ
s
p
nTun
h
n
h
a
n
T
A
1
s
1
n
T
A
2
s
2
n
T
...
A
p
s
p
n
T
u
n
Since
A
k
ⅇ
s
k
Tnun≡
A
k
zz-ⅇ
s
k
T
A
k
s
k
T
n
u
n
A
k
z
z
s
k
T
where
|z|>|ⅇ
s
k
T|
z
s
k
T
, and
Hz=∑k=1p
A
k
zz-ⅇ
s
k
T
H
z
k
1
p
A
k
z
z
s
k
T
where
|z|>maxk{|ⅇ
s
k
T|}
z
k
s
k
T
.
This technique is used occasionally in digital simulations of analog filters.
Problem 1What is the main
problem/drawback with this design technique?
[
Click for Solution 1 ]
Solution 1Since it samples the non-bandlimited
impulse response of the analog prototype filter, the frequency
response aliases. This distorts the
original analog frequency and destroys any optimal frequency
properties in the resulting digital filter.
[
Hide Solution 1 ]